--- title: "Recovery Discount in Dunning — Retention Growth Tactic" description: "Late in the dunning sequence, offer a discount on the next cycle as the final recovery lever—platform data across 1,029 orgs shows discounts help at 55 percent of orgs and hurt at 19 percent, so measure before standardizing the play." canonical: "https://churnkey.co/growth/library/recovery-discount-in-dunning" category: "growth" related: - /growth/library/index.html.md - /growth/library/methodology.html.md - /feature/payment-recovery/dunning-campaigns/index.html.md last_updated: "2026-06-09" --- # Recovery Discount in Dunning Late in the dunning sequence, offer a discount on the next cycle as the final recovery lever—platform data across 1,029 orgs shows discounts help at 55 percent of orgs and hurt at 19 percent, so measure before standardizing the play. - **Stage:** Retention - **Evidence grade:** strong (large, consistent correlation) - **Evidence:** Discounts help at 55 percent of orgs, are neutral at 26 percent, and hurt at 19 percent—measure before you standardize - **Trigger type:** Scheduled - **Channels:** email ## What is it? Most failed payments are mechanical—an expired card, a hit limit—but a minority sit on a different fault line: the card works, and the conviction wobbled. The subscriber saw the failure notice and treated it as a decision point rather than a chore. For that subscriber, a concrete saving on the next cycle tips the update. The tactic places a discount offer late in the dunning email sequence, after the mechanical-failure population has already been recovered by the standard attempts. It is deliberately the last lever, not the first: discounting a subscriber who only needed a working card gives margin away for nothing. ## What the evidence shows Platform discount analysis across 1,029 orgs shows discounts help at 55 percent of orgs, are neutral at 26 percent, and actively hurt at 19 percent. The tactic ships with its own warning label: at roughly one org in five, the discount trains subscribers to wait for offers or signals doubt about the price, and retention ends up worse than no offer at all. The heterogeneity is the finding. There is no universal answer to whether dunning discounts work—there is only an org-level answer, and the dataset is large enough to make checking it a prerequisite rather than a refinement. The org measures its own discount profile first, then standardizes or disables accordingly. ## How it runs The discount step sits late in the scheduled email sequence, firing only after at least two non-discount recovery attempts have run. By that point the remaining population skews toward conviction failures rather than card failures, which is exactly the population a saving can move. The offer itself stays narrow: it applies to the next billing cycle only and never silently recurs, it never stacks with another active promotion, and it is switched off entirely at orgs whose discount-effectiveness profile says discounts hurt. The lever is real, but it is a scalpel with a pre-check, not a default. ## Run it with the Churnkey MCP This tactic runs via the Churnkey MCP. Connect it to your favorite AI agent and it reads your own usage and billing data, then recommends and runs the plays most likely to move your LTV — starting with whether this one fits. **Trigger:** A late scheduled step in the dunning sequence fires for a subscriber still unrecovered after the standard non-discount attempts. **Surface:** A discount offer delivered in the final emails of the dunning sequence, applied to the next billing cycle on payment update. **Guardrails:** - Position: the discount fires late in the sequence, after at least two non-discount recovery attempts—never as the opening move. - Pre-check: verify the org’s own discount-effectiveness profile first and disable the step at orgs where discounts hurt. - Scope: the discount applies to the next cycle only, never silently recurring—the saving is a bridge, not a new price. - Exclusivity: never stack with any other active promotion; one concession per recovery, clearly stated. **Evidence:** Platform discount analysis across 1,029 orgs shows discounts help at 55 percent of orgs, are neutral at 26 percent, and hurt at 19 percent. The org-level heterogeneity is the finding—the discount only earns its place at orgs where the data says discounts help. **How to run it:** 1. Measure the org’s discount-effectiveness profile from its own offer history before enabling the step. 2. Place the discount email late in the dunning schedule, gated on at least two prior non-discount attempts. 3. Configure the offer as next-cycle-only with automatic expiry, and block stacking with any other active promotion. 4. Apply the discount automatically when the subscriber updates payment through the recovery link. 5. Compare recovery rate and downstream retention for the discount step against the non-discount baseline, and disable it if the org’s data turns negative. Install the MCP server: ```bash npm install -g @churnkey/mcp ``` The full reference is in the [Churnkey MCP docs](https://docs.churnkey.co/data-integrations/mcp). ## Related - [Growth Tactics Library](/growth/library/index.html.md) — The full register of graded growth plays. - [How Tactics Are Graded](/growth/library/methodology.html.md) — The four-grade evidence ladder behind this grade. - [Dunning Campaigns](/feature/payment-recovery/dunning-campaigns/index.html.md) — The Churnkey feature this tactic maps to in production.